Dementia classification using two-channel electroencephalography features.

Area of Science:

• Neuroscience
• Biomedical Engineering

Background:

• Dementia diagnosis relies on clinical assessments and neuroimaging.
• Wearable electroencephalography (EEG) offers a potential non-invasive method for monitoring brain activity.

Purpose of the Study:

• To develop and validate a novel classification model for differentiating dementia patients from normal controls (NCs) using wearable two-channel EEG data.
• To identify key EEG features indicative of dementia.

Main Methods:

• Utilized an extreme gradient boosting (Xgboost) model combined with recursive feature elimination with cross-validation (RFECV).
• Recorded resting-state EEG data from 54 NCs and 29 dementia patients.
• Performed Mini-Mental Status Exam (MMSE) and Clinical Dementia Rating (CDR) assessments.

Main Results:

• Significant differences in peak frequency (PF), alpha (A), theta (T), A/T ratio, A/BL ratio, and coherence (CH) were observed.
• Dementia patients showed decreased PF, CH_A/T, CH_A/BL, A/T, and A/BL, with increased T.
• The Xgboost model with RFECV achieved a balanced accuracy of 97.05%, with PF as the most discriminative feature.

Conclusions:

• The developed Xgboost with RFECV model effectively differentiates dementia patients from NCs using two-channel EEG.
• This novel approach demonstrates the potential of wearable EEG for accessible and accurate dementia diagnosis.